Add a first-class configurational expression system

[DiogoRibeiro7]

Replaces string handling of QCA notation with a typed tree: tokenizer, recursive-descent parser, AST, canonical form and evaluation.

Design

Built on the existing sets.py nodes rather than beside them, so Condition/Negation/Intersection/Union remain the node types and nothing is duplicated. Conjunction and Disjunction are exported as the names the configurational literature uses. New nodes: Implication and Configuration.

parse_expression("A*~B + C")     # tree
parse_expression("A*B -> Y")     # Implication
evaluate_expression("A*~B", df)  # membership
simplify_expression("A + A*B")   # A

Two things worth review

A real bug fixed. The old printer dropped grouping: (A+B)*C rendered as A+B*C, which re-parses as A+(B*C) — a different set. The printer now re-inserts parentheses by precedence, and round-tripping is property-tested.

The complement laws are deliberately not applied. In Boolean algebra A*~A is empty; for fuzzy sets it is not — at A = 0.5, min(A, 1-A) = 0.5. Simplification is restricted to laws that hold for min/max/1−x: associativity, commutativity, idempotence, double negation and absorption. A test pins A*~A and A+~A as irreducible, and a property test asserts simplification never changes membership on real data.

Safety

Parsing is structural — no eval, no exec. An expression from a config file or user input cannot execute anything; there is a test asserting that an __import__ payload is a syntax error. Errors carry the offending position.

Tests

45 new tests: tokenizer, precedence, parenthesisation, implications, malformed input, canonical form, absorption, configurations and minterm round-trips, plus two Hypothesis properties — text → tree → text → tree stays semantically identical, and simplification preserves membership.

225 tests pass at 100% coverage; mypy strict and docs build clean.

---

Summary by cubic

Adds a first-class configurational expression system that parses QCA notation into a typed tree with evaluation, simplification, and safe, correct printing. This replaces string handling, improves safety, and makes expressions comparable and testable.

• New Features

  • Parse QCA expressions into a typed AST (*, +, ~, ->), reusing Condition/Negation/Intersection/Union; new nodes: Implication, Configuration (also export Conjunction/Disjunction).
  • Evaluate memberships with evaluate_expression(...); evaluate sufficiency claims via Implication.evaluate_relation(...).
  • Simplify using fuzzy-safe laws (associativity, commutativity, idempotence, double negation, absorption) with simplify_expression(...); provide canonical form and equivalent(...). Complement laws are intentionally not applied.
  • Structural parsing only; no eval. Errors raise ExpressionSyntaxError with precise positions.
  • Configuration helpers: Configuration.from_minterm(...), .to_expression(), big-endian minterm mapping.
  • Export the API from setqca (parse_expression, evaluate_expression, simplify_expression, Implication, Configuration, etc.). Added guide at docs/guide/expressions.md and extensive property tests.

• Bug Fixes

  • Fixed printer losing parentheses: (A+B)*C no longer renders as A+B*C. Printing now respects precedence and round-trips correctly.

^{Written for commit 7478c9e. Summary will update on new commits.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!